Semi-supervised Laplacianfaces from Pairwise Constraints for Face Recognition

Adaptive semi-supervised dimensionality reduction with sparse representation using pairwise constraints

We consider the problem of deep semi-supervised classification, where label information is obtained in the form of pairwise constraints. Existing approaches to this problem begin with a clustering network and utilize custom loss functions to encourage the learned representations to conform to the obtained constraints. We present a novel framework that seamlessly integrates pairwise constrained clustering, semi-supervised classification, and supervised classification. This approach leverages advances in unsupervised learning by jointly training a Siamese network and autoencoder to learn a representation that is amenable for both clustering and classification. The resulting framework outperforms existing approaches on common image recognition datasets.

While in standard clustering no side information is used, users might be interested in providing additional information to influence the clustering. In case of document clustering, additional information can take the form of pairwise constraints where a user provides additional information about pairs of documents as must-link and cannot-link constraints (indicating respectively whether the documents in the pair are coming from the same cluster or not). In this paper, we propose a novel deep document clustering framework which can employ pairwise constraints while learning document representations to obtain better tailored results. Indeed, in our proposed framework, data representations (obtained through an autoencoder) and cluster representatives are learned through back propagation in a joint way. Devising a fully differentiable deep clustering framework with the ability of using pairwise constraints is the main contribution of this paper. Experiments conducted on 5 public datasets show the gain in clustering performance which the resulting approach can yield.

As a variant of nonnegative matrix factorization (NMF), symmetric NMF (SNMF) has shown to be effective for capturing the cluster structure embedded in the graph representation. In contrast to the existing SNMF-based clustering methods that empirically construct the similarity matrix and rigidly introduce the supervisory information to the assignment matrix, in this paper, we propose a novel SNMF-based semisupervised clustering method, namely, pairwise constraint propagation-induced SNMF (PCPSNMF). By formulating a single-constrained optimization problem, PCPSNMF is capable of learning the similarity and assignment matrices adaptively and simultaneously, in which a small amount of supervisory information in the form of pairwise constraints is introduced in a flexible way to guide the construction of the similarity matrix, and the two matrices communicate with each other to achieve mutual refinement until convergence. In addition, we propose an efficient alternating iterative algorithm to solve the optimization problem, whose convergence is theoretically proven. Experimental results over several benchmark image data sets demonstrate that PCPSNMF is less sensitive to initialization and produces higher clustering performance, compared with the state-of-the-art methods.

Recently, how to exploit unlabeled data for training face recognition models has been attracting increasing attention. However, few works consider the unlabeled shallow data <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> in real-world scenarios. The existing semi-supervised face recognition methods that focus on generating pseudo labels or minimizing softmax classification probabilities of the unlabeled data do not work very well on the unlabeled shallow data. It is still a challenge on how to effectively utilize the unlabeled shallow face data to improve the performance of face recognition. In this paper, we propose a novel face recognition method, named VirFace, to effectively exploit the unlabeled shallow data for face recognition. VirFace consists of VirClass and VirInstance. Specifically, VirClass enlarges the inter-class distance by injecting the unlabeled data as new identities, while VirInstance produces virtual instances sampled from the learned distribution of each identity to further enlarge the inter-class distance. To the best of our knowledge, we are the first to tackle the problem of unlabeled shallow face data. Extensive experiments have been conducted on both the small- and large-scale datasets, e.g. LFW and IJB-C, etc, demonstrating the superiority of the proposed method.

The research on community structure is a key to analyze the network functionality and topology, and thus it is significant to detect and analysis the community structure. During the abstract process from an actual system to a network, especially for a large-scale network, it is inevitable to have mistaken connections between nodes or have connection missing. In addition, in real applications, from time to time we can obtain prior information in the form of pairwise constraints between nodes besides topology information, although they may be inaccurate or conflicted. These noises in the network-related information will dramatically reduce the accuracy of community detection. Hence, in this paper, we introduce a dissimilarity index to determine the trustworthiness of pairwise constraints and settle the conflict of pairwise constraints. Then, focusing on the community detection with false connections or conflicted connections, we propose a pairwise constrained structure-enhanced extremal optimization-based semi-supervised algorithm (PCSEO-SS algorithm). Compared with existing semi-supervised community detection approaches, the experimental results executed on real networks and synthetic networks, show that PCSEO-SS can solve the problem of false connections or conflicted connections to some extent and detect the community structure more precisely.

Improving clustering with constrained communities

SSC-EKE: Semi-supervised classification with extensive knowledge exploitation

Semi-supervised classification with pairwise constraints: A case study on animal identification from video

Cayley-Klein metric is a specific kind of non-Euclidean metric in projective space. Recently, it has been introduced into metric learning with encouraging performance when dealing with computer vision tasks. However, the original Cayley-Klein metric learning methods with conventional pairwise and triplet-wise constraints, which are fixed bound based constraints, may not perform well when the intra-and inter-class variations of data distribution become complex. Pairwise constraints restrict the distance between samples of a similar pair to be lower than a fixed upper bound, and the distance between samples of a dissimilar pair higher than a fixed lower bound. Triplet-wise constraints restrict the distance between samples of a similar pair to be smaller than that between a pair of samples from different classes. In this paper, we propose a novel Cayley-Klein metric learning method (CKseML) with adaptive shrinkage-expansion pairwise constraints. CKseML is very effective in learning metric from data with complex distributions. Our experimental results demonstrate that CKseML achieves better performance than the original Cayley-Klein metric learning methods.

This paper proposes an unsupervised classification method for multilook polarimetric synthetic aperture radar (Pol-SAR) data. The proposed method simultaneously deals with the heterogeneity and incorporates the local correlation in PolSAR images. Specifically, within the probabilistic framework of the Dirichlet process mixture model (DPMM), an observed PolSAR data point is described by the multiplication of a Wishartdistributed component and a class-dependent random variable (i.e., the textual variable). This modeling scheme leads to the proposed textured DPMM (tDPMM), which possesses more flexibility in characterizing PolSAR data in heterogeneous areas and from high-resolution images due to the introduction of the classdependent texture variable. The proposed tDPMM is learned by solving an optimization problem to achieve its Bayesian inference. With the knowledge of this optimization-based learning, the local correlation is incorporated through the pairwise constraint, which integrates an appropriate penalty term into the objective function so as to encourage the neighboring pixels to fall into the same category and to alleviate the "salt-and-pepper" classification appearance.We develop the learning algorithm with all the closed-form updates. The performance of the proposed method is evaluated with both low-resolution and high-resolution PolSAR images, which involve homogeneous, heterogeneous, and extremely heterogeneous areas. The experimental results reveal that the class-dependent texture variable is beneficial to PolSAR image classification and the pairwise constraint can effectively incorporate the local correlation in PolSAR images.

Existing algorithms aiming to learn a binary classifier from positive (P) and unlabeled (U) data generally require estimating the class prior or label noises ahead of building a classification model. However, the estimation and classifier learning are normally conducted in a pipeline instead of being jointly optimized. In this paper, we propose to alternatively train the two steps using reinforcement learning. Our proposal adopts a policy network to adaptively make assumptions on the labels of unlabeled data, while a classifier is built upon the output of the policy network and provides rewards to learn a better strategy. The dynamic and interactive training between the policy maker and the classifier can exploit the unlabeled data in a more effective manner and yield a significant improvement on the classification performance. Furthermore, we present two different approaches to represent the actions sampled from the policy. The first approach considers continuous actions as soft labels, while the other uses discrete actions as hard assignment of labels for unlabeled examples. We validate the effectiveness of the proposed method on two benchmark datasets as well as one e-commerce dataset. The result shows the proposed method is able to consistently outperform state-of-the-art methods in various settings.

While recent studies on semi-supervised learning have shown remarkable progress in leveraging both labeled and unlabeled data, most of them presume a basic setting of the model is randomly initialized. In this work, we consider semi-supervised learning and transfer learning jointly, leading to a more practical and competitive paradigm that can utilize both powerful pre-trained models from source domain as well as labeled/unlabeled data in the target domain. To better exploit the value of both pre-trained weights and unlabeled target examples, we introduce adaptive consistency regularization that consists of two complementary components: Adaptive Knowledge Consistency (AKC) on the examples between the source and target model, and Adaptive Representation Consistency (ARC) on the target model between labeled and unlabeled examples. Examples involved in the consistency regularization are adaptively selected according to their potential contributions to the target task. We conduct extensive experiments on popular benchmarks including CIFAR-10, CUB-200, and MURA, by fine-tuning the ImageNet pre-trained ResNet-50 model. Results show that our proposed adaptive consistency regularization outperforms state-of-the-art semi-supervised learning techniques such as Pseudo Label, Mean Teacher, and FixMatch. Moreover, our algorithm is orthogonal to existing methods and thus able to gain additional improvements on top of MixMatch and FixMatch. Our code is available at https://github.com/Walleclipse/Semi-Supervised-Transfer-Learning-Paddle.

Scalable semi-supervised clustering by spectral kernel learning

Unlabeled training examples are readily available in many applications, but labeled examples are fairly expensive to obtain. For instance, in our previous works on classification of peer-to-peer (P2P) Internet traffics, we observed that only about 25% of examples can be labeled as “P2P”or “NonP2P” using a port-based heuristic rule. We also expect that even fewer examples can be labeled in the future as more and more P2P applications use dynamic ports. This fact motivates us to investigate the techniques which enhance the accuracy of P2P traffic classification by exploiting the unlabeled examples. In addition, the Internet data flows dynamically in large volumes (streaming data). In P2P applications, new communities of peers often join and old communities of peers often leave, requiring the classifiers to be capable of updating the model incrementally, and dealing with concept drift. Based on these requirements, this paper proposes an incremental Tri-Training (iTT) algorithm. We tested our approach on a real data stream with 7.2 Mega labeled examples and 20.4 Mega unlabeled examples. The results show that iTT algorithm can enhance accuracy of P2P traffic classification by exploiting unlabeled examples. In addition, it can effectively deal with dynamic nature of streaming data to detect the changes in communities of peers. We extracted attributes only from the IP layer, eliminating the privacy concern associated with the techniques that use deep packet inspection.